AuriAetherwiing/MN-12B-Starcannon-v2

Hugging Face
TEXT GENERATIONConcurrency Cost:1Model Size:12BQuant:FP8Ctx Length:32kPublished:Aug 1, 2024License:cc-by-nc-nd-4.0Architecture:Transformer0.0K Open Weights Warm

AuriAetherwiing/MN-12B-Starcannon-v2 is a 12 billion parameter merged language model, created using the TIES merge method from nothingiisreal/MN-12B-Celeste-V1.9 and intervitens/mini-magnum-12b-v1.1. This model exhibits a creative and verbose writing style, offering more variety than Magnum and more detailed prose than Celeste v1.9. It is designed for users seeking a distinct narrative voice and enhanced creative text generation.

Loading preview...

MN-12B-Starcannon-v2 Overview

MN-12B-Starcannon-v2 is a 12 billion parameter language model developed by AuriAetherwiing, created through a merge of existing pre-trained models using the mergekit tool. This model specifically utilizes the TIES merge method, combining the characteristics of nothingiisreal/MN-12B-Celeste-V1.9 as its base and intervitens/mini-magnum-12b-v1.1.

Key Characteristics

  • Creative Writing Style: The model is noted for its creative output and verbose prose, offering a distinct narrative voice.
  • Blended Personalities: It combines elements from both Celeste v1.9 and mini-magnum-12b-v1.1, aiming for a balance that provides more variety than Magnum and more detailed writing than Celeste v1.9.
  • Merge Method: Built using the TIES (Task-Independent Ensemble of Subnetworks) merge method, which allows for combining multiple models effectively.

Use Cases

This model is particularly suited for applications requiring:

  • Creative Text Generation: Ideal for scenarios where unique and imaginative text is desired.
  • Narrative Development: Its verbose and varied writing style can be beneficial for generating stories, roleplay scenarios, or descriptive content.

Quantized versions are available, including Dynamic FP8, Static GGUF, and EXL2 formats.

Popular Sampler Settings

Top 3 parameter combinations used by Featherless users for this model. Click a tab to see each config.

temperature
top_p
top_k
frequency_penalty
presence_penalty
repetition_penalty
min_p